Quick power supply switching auxiliary decision method based on digitalized preplan and topology deduction

Abstract

The application discloses a quick power transfer auxiliary decision-making method based on digital plan and topology deduction, and belongs to the field of power grid dispatching control, which comprises the following steps: collecting real-time data, predicting data and identifying fault of flexible resource state; fusing data to construct a power grid dynamic situation map and instantiate a digital twin sand table; obtaining strategy gene operators in a digital plan library and recombining to generate an initial strategy pedigree tree; using a time-space composite stress field to deduce and prune, combining with a flexible adjustment operator to generate a composite strategy set; executing antagonistic deduction in the sand table to obtain a risk migration trajectory and a robustness index; and applying a multi-level game evaluation model to output a recommended strategy. The application adopts strategy gene recombination and digital twin antagonistic deduction technology, can realize automatic generation and safety deep checking of a power recovery scheme, and improves the scientificity and execution efficiency of the power transfer decision under heavy load working conditions and multi-factor interference.

Description

Technical Field

[0001] This invention relates to the field of power grid dispatch and control, and in particular to a rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation. Background Technology

[0002] Power transfer is a crucial dispatching operation in power systems. It refers to the process by which dispatchers, after a power grid equipment failure causes a partial power outage, alter the grid topology by opening or closing a series of switches to allow other healthy power sources or lines to restore power to the affected areas. This operation plays a vital role in ensuring the reliability of power supply, minimizing power outage time, and maintaining socio-economic stability. Rapid and accurate power transfer decisions are a core requirement for the safe and stable operation of modern power grids, especially increasingly complex distribution networks.

[0003] In existing technologies, power transfer decisions after a power grid failure primarily rely on pre-established contingency plans and the personal experience of dispatchers. These plans are typically designed for several typical, high-probability failure scenarios and are stored in text or flowchart form. When a failure occurs, dispatchers first need to determine the failure type, then search for a matching plan in the plan library, and combine this with basic information such as the current power flow to assess feasibility, ultimately manually generating an operational sequence. For atypical failures not covered by the plans, the dispatcher's on-site analysis and handling capabilities become even more crucial.

[0004] However, the aforementioned existing technical solutions have significant drawbacks. Static emergency response plans are ill-suited to the real-time dynamic changes in the power grid. Grid load levels, renewable energy output, and topology are all constantly changing, and the operational paths defined in the plans may become inapplicable due to line overload or voltage exceeding limits during actual fault occurrences. Traditional decision-making processes lack forward-looking risk assessment; strategies based solely on current cross-sectional information may become unsafe within tens of minutes due to natural load increases. Existing methods fail to effectively integrate new flexible resources such as energy storage and controllable loads, limiting the means and potential for fault recovery. The decision-making process heavily relies on manual intervention, resulting in low efficiency and a high risk of error, making it difficult to quickly find the globally optimal solution among numerous possible recovery paths. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation. It employs a digital twin sandbox constructed by integrating multi-source data, dynamically reorganizes and evolves strategies, and uses spatiotemporal stress fields and adversarial simulations for multi-dimensional evaluation. This enables the safe and intelligent generation of power transfer strategies after grid faults, thereby improving the reliability and resilience of grid operation.

[0006] The above objectives can be achieved through the following approach:

[0007] A rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation includes: collecting real-time power grid operation data, environmental prediction data, and flexible resource status data; identifying power grid fault information based on the real-time operation data; fusing the power grid fault information, the environmental prediction data, and the flexible resource status data to construct a dynamic power grid situation map, including power grid topology connections, real-time and future predicted carrying capacity parameters of each path, and callable flexible adjustment capacity parameters of each node; and constructing a digital twin sandbox that operates synchronously with the physical power grid in a digital twin environment based on the dynamic power grid situation map; obtaining strategy gene operators that match the power grid fault information from a digital contingency plan library constructed based on text contingency plan parsing, and then applying these operators according to the characteristics of the dynamic power grid situation map. The system employs a dynamic recombination strategy gene operator to generate an initial strategy phylogenetic tree. It maps the physical security constraints in the power grid dynamic situation diagram to a spatiotemporal composite stress field, uses this stress field to deduce and prune the initial strategy phylogenetic tree, and calls flexible adjustment operators stored in a flexible resource operator library to generate an evolved composite strategy set. In a digital twin sandbox, it creates parallel deduction instances for each composite strategy in the set, injects interference factors provided by an interference factor library for adversarial deduction, and obtains the risk migration trajectory and robustness indicators of each composite strategy. Finally, it applies a pre-defined multi-level game evaluation model containing security, efficiency, and resilience layers to comprehensively evaluate each composite strategy that has completed adversarial deduction, and outputs recommended strategies based on the evaluation results.

[0008] Optionally, by integrating the power grid fault information, the environmental prediction data, and the flexible resource status data, a dynamic power grid situation map is constructed, which includes the power grid topology connection relationship, the real-time and future predicted carrying capacity parameters of each path, and the callable flexible adjustment capacity parameters of each node. This includes: generating a current power grid topology connection map based on the power grid topology and power flow information in the real-time operating data; mapping the ultra-short-term load prediction curve and the new energy power generation prediction curve in the environmental prediction data to the relevant equipment and paths in the current power grid topology connection map, and calculating the predicted carrying capacity parameters for several predetermined periods in the future; obtaining flexible resource status data, and marking the adjustable capacity parameters of each flexible resource on the corresponding nodes of the current power grid topology connection map to form a dynamic power grid situation map.

[0009] Optionally, based on the characteristics of the power grid dynamic situation diagram, dynamically reorganizing the strategy gene operators to generate an initial strategy genealogy tree containing several recovery path sequences includes: performing natural language processing on the text plan to parse out a structured strategy gene operator consisting of triggering conditions, execution subjects, action instructions, and safety boundaries, and storing it in a digital plan library; indexing the basic operation operator sequence from the digital plan library based on the fault equipment identifier and outage range in the power grid fault information; using the basic operation operator sequence as the root node, and based on the real-time carrying capacity parameters and connection relationships of each path in the power grid dynamic situation diagram, performing logical equivalent replacement and expansion on the operation objects and order in the basic operation operator sequence to form several branches, constituting the initial strategy genealogy tree.

[0010] Optionally, using the spatiotemporal composite stress field to deduce and prune the initial strategy tree includes: quantifying the overload risk of the line and the voltage limit exceedance risk of the node into spatial stress weights, and quantifying the load increase pressure and new energy fluctuation pressure in the future predetermined period into temporal stress weights, which together constitute the spatiotemporal composite stress field; sequentially performing simulation operations on each recovery path sequence in the initial strategy tree, and calculating the distribution change of the spatiotemporal composite stress field in the power grid dynamic situation diagram after each simulation operation; according to a preset safety threshold, if a certain simulation operation causes the stress value at any position in the spatiotemporal composite stress field to exceed the safety threshold, then pruning the path branch containing the simulation operation.

[0011] Optionally, calling flexible adjustment operators stored in the flexible resource operator library to generate the evolved composite strategy set includes: identifying critical overload paths or voltage-weak nodes that lead to pruning during the deduction and pruning of the initial strategy phylogenetic tree; retrieving target flexible adjustment operators from the flexible resource operator library that can alleviate the pressure on the critical overload paths or voltage-weak nodes, wherein the target flexible adjustment operators include energy storage discharge operators and controllable load reduction operators; inserting the target flexible adjustment operators after the simulation operation steps corresponding to the critical overload paths or voltage-weak nodes to generate a composite strategy that integrates network reconfiguration operations and flexible resource adjustment operations, thus generating the evolved composite strategy set.

[0012] Optionally, the adversarial simulation by injecting interference factors provided by the interference factor library includes: performing simulated operations in the digital twin sandbox according to the operation sequence and parameters of the composite strategy; automatically injecting several interference factors simulating instantaneous load changes, unexpected power source shutdowns, or communication delays into the simulation instance at preset key operation time points or equipment state change points; monitoring and recording the changes in the power grid state parameters of the simulation instance under the influence of the interference factors; if equipment exceeds limits, determining that the simulation instance has failed under the interference factors, and recording the cause of failure and risk migration trajectory; and calculating the robustness index based on the success rate and state deviation of the simulation instance under different interference factor injections.

[0013] Optionally, the method further includes: collecting the deduction process data and result data of all the deduction instances to form a strategy deduction case library; analyzing the strategy gene combination patterns and corresponding power grid dynamic situation map characteristics of successful cases in the strategy deduction case library, and strengthening the correlation weight of the strategy gene operator in the digital contingency plan library; analyzing the failure causes of failed cases, and extracting the failure causes into new constraints for optimizing the calculation model of the spatiotemporal composite stress field or enriching the interference factor library.

[0014] Optionally, the application uses a pre-defined multi-level game evaluation model comprising a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate each composite strategy that completes the adversarial simulation. This evaluation includes: a first-level security layer evaluation, which selects composite strategies that do not cause any equipment parameter exceedances during the adversarial simulation, forming a set of security strategies; a second-level efficiency layer evaluation, which calculates the power restoration ratio and the total number of operation steps for each composite strategy within the set of security strategies, and performs an efficiency trade-off; a third-level resilience layer evaluation, which assesses the resilience of candidate strategies after the efficiency layer evaluation based on the robustness index and the dispersion of the risk migration trajectory; and finally, by combining the results of the security layer evaluation, efficiency layer evaluation, and resilience layer evaluation, a Pareto optimal frontier method is used to select a recommended strategy.

[0015] Optionally, after selecting the recommended strategy, the method further includes: converting the operation sequence and flexible adjustment instructions corresponding to the recommended strategy into a structured operation ticket that conforms to the scheduling procedure; and visually displaying the structured operation ticket, the risk migration trajectory, and key early warning information.

[0016] Based on the same inventive concept, this invention also provides a rapid power transfer auxiliary decision-making system based on digital contingency plans and topology simulation. The system includes: a data acquisition and identification module for acquiring real-time power grid operation data, environmental prediction data, and flexible resource status data, and identifying power grid fault information based on the real-time operation data; a situation construction and sandbox instantiation module for fusing the power grid fault information, the environmental prediction data, and the flexible resource status data to construct a dynamic power grid situation map containing power grid topology connections, real-time and future predicted carrying capacity parameters of each path, and callable flexible adjustment capacity parameters of each node, and constructing a digital twin sandbox that operates synchronously with the physical power grid in a digital twin environment based on the dynamic power grid situation map; and a contingency plan reorganization and phylogenetic tree generation module for obtaining strategy gene operators matching the power grid fault information from a digital contingency plan library constructed based on text contingency plan parsing, and generating a phylogenetic tree based on the power grid... The system features a dynamic situational awareness map, a dynamic recombination strategy gene operator to generate an initial strategy phylogenetic tree; a stress field deduction and composite evolution module to map the physical security constraints in the power grid dynamic situational awareness map into a spatiotemporal composite stress field, use the spatiotemporal composite stress field to deduce and prune the initial strategy phylogenetic tree, and call flexible adjustment operators stored in the flexible resource operator library to generate an evolved composite strategy set; a sandbox adversarial deduction and index acquisition module to create parallel deduction instances for each composite strategy in the composite strategy set in the digital twin sandbox, inject interference factors provided by the interference factor library to conduct adversarial deduction, and obtain the risk migration trajectory and robustness index of each composite strategy; and a game evaluation and strategy recommendation module to apply a preset multi-level game evaluation model containing a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate each composite strategy that has completed adversarial deduction, and output recommended strategies based on the evaluation results.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] This invention provides a rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction. By dynamically constructing a dynamic power grid situation map that integrates real-time, predictive, and resource status data, it achieves comprehensive and accurate perception of the power grid status. This method can transform static text contingency plans into dynamic and diverse initial strategy phylogenetic trees based on real-time power grid characteristics, improving the adaptability and flexibility of strategy generation and overcoming the shortcomings of traditional contingency plans that are rigid and unable to adapt to changing operating conditions.

[0019] This invention introduces a spatiotemporal composite stress field to rapidly verify the physical security of a strategy, and combines adversarial simulation to evaluate the strategy's robustness under uncertain disturbances, constructing a multi-layered, in-depth defense assessment system from static security to dynamic resilience. This method can expose and avoid potential risks that are difficult to detect using traditional methods in advance, ensuring that the ultimately recommended strategy is not only feasible at present but also capable of withstanding unknown future risks, thereby improving the overall security and reliability of power grid transfer operations.

[0020] This invention seamlessly integrates flexible resources such as energy storage and controllable loads into the power transfer decision-making process in the form of operators, achieving deep synergy between traditional network reconfiguration and new resource regulation. When bottlenecks exist in traditional power transfer paths, flexible resources can be proactively invoked for peak shaving and valley filling or voltage support, thereby repairing and evolving strategies and creating new feasible recovery paths. This not only expands the means of fault recovery but also enhances the grid's operational resilience and adaptability to fluctuations in renewable energy sources.

[0021] This invention employs a multi-level game theory evaluation model comprising a security layer, an efficiency layer, and a resilience layer, and uses the Pareto optimal frontier method for final decision recommendation. This model can systematically weigh multiple conflicting objectives, providing schedulers with a set of overall optimal strategy options, rather than a single suboptimal solution. This makes the decision-making process more scientific and transparent, meeting the differentiated needs for recovery speed, scope, and safety margins in various scenarios, and achieving refined and optimized decision-making. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction, according to an embodiment of the present invention.

[0024] Figure 2 This is a stress evolution trend diagram of the transfer path in the operation simulation process according to an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of the structure of the rapid power transfer auxiliary decision-making system based on digital contingency plans and topology deduction according to an embodiment of the present invention.

[0026] Figure 4 This is the optimal strategy selection space graph under the multi-level game evaluation model of this invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1 One embodiment of the present invention proposes a rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation. It adopts a digital twin sandbox constructed by fusing multi-source data, dynamically reorganizes and evolves strategies, and conducts multi-dimensional evaluation through spatiotemporal stress fields and adversarial simulations. This method can realize the rapid, safe and intelligent generation of power transfer strategies after grid failures, and significantly improve the reliability and resilience of grid operation.

[0029] The method described in this embodiment specifically includes:

[0030] Collect real-time operation data of the power grid, environmental prediction data, and flexible resource status data, and identify power grid fault information based on the real-time operation data;

[0031] By integrating the power grid fault information, the environmental prediction data, and the flexible resource status data, a dynamic power grid situation map is constructed, which includes the power grid topology connection relationship, the real-time and future predicted carrying capacity parameters of each path, and the callable flexible adjustment capacity parameters of each node. Based on the dynamic power grid situation map, a digital twin sandbox that runs synchronously with the physical power grid is constructed in the digital twin environment.

[0032] The strategy gene operator that matches the power grid fault information is obtained from the digital contingency plan library constructed based on text contingency plan parsing, and the strategy gene operator is dynamically recombined according to the characteristics of the power grid dynamic situation map to generate an initial strategy genealogy tree.

[0033] The physical security constraints in the power grid dynamic situation diagram are mapped to a spatiotemporal composite stress field. The spatiotemporal composite stress field is used to deduce and prune the initial strategy phylogenetic tree, and the flexible adjustment operators stored in the flexible resource operator library are called to generate the evolved composite strategy set.

[0034] In the digital twin sandbox, a parallel simulation instance is created for each composite strategy in the composite strategy set, and interference factors provided by the interference factor library are injected to conduct adversarial simulation to obtain the risk migration trajectory and robustness index of each composite strategy.

[0035] A pre-defined multi-level game evaluation model, including a security layer, an efficiency layer, and a resilience layer, is used to comprehensively evaluate the various composite strategies that complete the adversarial simulation, and recommended strategies are output based on the evaluation results.

[0036] Specifically, a closed-loop intelligent process is constructed, encompassing data perception, strategy generation, risk extrapolation, and optimization decision-making. By integrating multi-source heterogeneous data, including real-time, predictive, and flexible resource data, a dynamic state-potential diagram reflecting the current state, future trends, and controllable potential of the power grid is built and instantiated as a digital twin sandbox, providing a high-fidelity panoramic virtual environment for decision-making. Breaking away from the rigidity of traditional contingency plans, text-based plans are parsed into flexibly recombinable strategy gene operators, and diverse initial strategy phylogenetic trees are dynamically generated based on real-time power grid conditions, achieving scenario-adaptive strategy generation. The concept of a spatiotemporal composite stress field is innovatively introduced, transforming complex physical constraints into quantifiable stress distributions. Unsafe strategies are quickly filtered out through extrapolation pruning, and promising strategies are evolved by invoking flexible adjustment operators, achieving synergistic optimization of rigid topology operations and flexible resource regulation. Based on this, adversarial extrapolation is conducted by introducing interference factors into the digital twin sandbox, deeply evaluating the robustness of strategies under uncertain disturbances and obtaining risk migration trajectories. The application employs a multi-level game evaluation model encompassing three dimensions: security, efficiency, and resilience. This model performs a comprehensive Pareto optimal screening of candidate strategies, ensuring that the final recommended strategy is the comprehensive optimal solution under multiple objective trade-offs.

[0037] Optionally, by integrating the power grid fault information, the environmental prediction data, and the flexible resource status data, a dynamic power grid situation map is constructed, including the power grid topology connections, real-time and future predicted carrying capacity parameters of each path, and callable flexible adjustment capacity parameters of each node.

[0038] Based on the power grid topology and power flow information in the real-time operating data, a current power grid topology connection diagram is generated;

[0039] The ultra-short-term load forecast curve and the new energy power generation forecast curve in the environmental forecast data are mapped to the relevant equipment and paths in the current power grid topology connection diagram to calculate the predicted carrying capacity parameters for several predetermined periods in the future.

[0040] The status data of the flexible resources are obtained, and the adjustable capacity parameters of each flexible resource are marked on the corresponding nodes of the current power grid topology connection diagram to form a dynamic status diagram of the power grid.

[0041] Specifically, this involves constructing a baseline snapshot of the power grid's physical connections and operational status. Real-time telemetry and teleindication data for the entire network are acquired from the Power Management System (EMS) or Supervisory Control and Data Acquisition (SCADA) system. Telemetry and teleindication data, particularly the 0 or 1 states of circuit breakers and disconnectors, are used to generate an adjacency matrix or graph structure describing the power grid's physical connections, forming the current power grid topology. Telemetry data such as active power, reactive power, and node voltage of each line and transformer are then used as weights or attributes to label the corresponding branches and nodes in this topology, completing the current snapshot of the power grid's physical connections. A static electrical profile is created. To endow the static topology with dynamic predictive capabilities over time and anticipate future risks, environmental forecast data is retrieved, specifically ultra-short-term load forecast curves and renewable energy generation forecast curves at 15-minute intervals within the next 4 hours. For each load node or renewable energy station node in the topology connection diagram, the corresponding forecast data sequence is used as the power injection or outflow time-series boundary condition for that node. Based on this, for each predetermined time period, for example... , For example, multiple power flow calculations are performed on the calculated cross-sections. These calculations yield the predicted power flow for each critical path over several predetermined time periods in the future. And combined with the static thermal stability limit in the equipment ledger The predicted carrying capacity parameters and predicted load rate are calculated. The calculation formula is:

[0042] ;

[0043] in, For connecting nodes With nodes The path in the future Predicted load rate for a scheduled time period; The path was obtained through power flow calculation. Predicted active power for the time period; This is the rated transmission capacity of the path, a constant value. This parameter sequence is appended to the corresponding path in the topology graph. To label the grid's active regulation capability on the dynamic topology, available control methods are quantified. Flexible resource status data is accessed, sourced from the battery management system (BMS) of energy storage systems, load control systems of demand-side response aggregators, etc. Key adjustable capacity parameters of flexible resources are extracted; for example, for grid-connected energy storage stations, the parameter is the maximum dischargeable power under the current state of charge (SOC) constraint. With maximum rechargeable power For interruptible load clusters, the parameter is the total load that can be reduced. These dynamically updated parameters are precisely labeled on the corresponding geographical locations or electrical nodes in the current power grid topology connection diagram. This ultimately generates a dynamic power grid situation map that integrates multi-source heterogeneous data. This situation map is a data-rich time-varying graphical model. Its structure not only includes the power grid topology connection relationships representing physical connections, but also attaches a sequence of real-time and future predicted carrying capacity parameters reflecting current and future operational pressures to each path. Furthermore, it explicitly labels the callable flexible adjustment capacity parameters of each node with real-time call potential at key nodes. This highly integrated dynamic power grid situation map provides accurate, dynamic, and forward-looking digital objects for subsequent strategy generation, deduction, and evaluation. Figure 2 As shown, the changes in the spatiotemporal composite stress values ​​of different candidate power transfer paths during the simulation operation are displayed, intuitively presenting the quantitative impact of line load pressure and future trends on path safety.

[0044] Optionally, based on the characteristics of the power grid dynamic situation map, the strategy gene operator is dynamically recombined to generate an initial strategy phylogenetic tree containing several recovery path sequences, including:

[0045] Natural language processing is performed on the text plan to parse it into a structured strategy gene operator consisting of triggering conditions, execution subject, action instructions and security boundaries, and then stored in the digital plan library.

[0046] Based on the fault equipment identifier and power outage range in the power grid fault information, the basic operation operator sequence is indexed from the digital contingency plan database;

[0047] Using the basic operation operator sequence as the root node, and based on the real-time carrying capacity parameters and connection relationships of each path in the power grid dynamic situation diagram, the operation objects and sequences in the basic operation operator sequence are logically equivalently replaced and expanded to form several branches, thus constituting the initial strategy spectrum tree.

[0048] Specifically, to build a machine-readable and structured knowledge base for contingency plans, natural language processing (NLP) techniques are used to parse historically accumulated textual contingency plans offline. Named entity recognition (NER) technology is used to extract information such as "..." from unstructured contingency plan texts. The system includes actuators such as the "Sunshine Substation," "Main Transformer No. 2," and "Line 952 Switch," as well as action commands such as "Open," "Close," and "Activate." Simultaneously, through relation extraction (RE) technology, triggering conditions are identified and associated, such as... The fault caused the line to be overloaded, and safety boundaries, such as "the bus voltage after operation must not be lower than..." These parsed elements are assembled into structured strategy gene operators, with a standard data format of {trigger condition, execution subject, action instruction, safety boundary}. All parsed strategy gene operators are categorized, indexed, and stored in a digital contingency plan library, forming a standardized set of basic operation instructions that can be quickly retrieved. This allows for the rapid identification and generation of standard, unoptimized basic recovery paths based on actual fault information. When a power grid fault occurs, the key fault equipment identifiers and the directly affected power outage range are extracted based on the power grid fault information provided by the data acquisition and identification module. For example, the fault information might be "Yangguang Substation". The information "Line 952 switch tripped due to a fault, causing power loss in area A" was used as a search keyword. A matching query was performed in the digital contingency plan database to index the set of strategy gene operators most relevant to this fault scenario. These were then arranged according to the inherent logical order in the contingency plan to form a basic operation operator sequence. This sequence represents the standard recovery procedure most commonly used by dispatchers under typical operating conditions, such as {closing the backup line 954 switch, checking load transfer status, ...}. This basic operation operator sequence constitutes the root node for subsequent strategy evolution. To adaptively mutate and expand the standard recovery path based on the real-time operating status of the power grid, diverse candidate strategies are generated. The root of the initial strategy phylogenetic tree is constructed using the basic operation operator sequence as the initial backbone. Each strategy gene operator in this sequence is traversed. For each operator, especially power transfer operations involving network reconfiguration, the power grid dynamic situation map is queried. The real-time carrying capacity parameters of the "executing entity" in the operator are closely monitored. If the real-time load rate of the path exceeds a preset threshold, the standard path is deemed to be at risk, and a logical trigger is activated. The mechanism employs an equivalent replacement mechanism. Based on the topological connections in the power grid dynamic situation diagram, it uses graph search algorithms, such as the k-shortest path algorithm, to find other electrically feasible alternative paths from the same power source to the de-energized load area. For each alternative path that satisfies the basic capacity constraints, a new set of policy gene operators is generated to describe the actions performed on that path, replacing the risky operations in the original sequence, thus generating new branches on the policy tree. This process is repeated continuously, expanding a single standard operation sequence into a tree structure containing several recovery path sequences—the initial policy policy tree. Through these steps, a transformation from static knowledge to dynamic policy is completed. Each complete path from the root node to a leaf node in the output initial policy policy tree represents a theoretically feasible recovery path sequence under the current power grid topology. This tree not only includes standard contingency plans derived from expert experience but also derives a large number of variant policies adapted to the current power grid conditions, expanding the decision space and providing a solid foundation for subsequent deeper physical safety verification and optimization.

[0049] Optionally, using the spatiotemporal composite stress field to deduce and prune the initial strategy phylogenetic tree includes:

[0050] The overload risk of the line and the voltage over-limit risk of the node are quantified into spatial stress weights, and the load increase pressure and new energy fluctuation pressure in the future predetermined period are quantified into temporal stress weights, which together constitute a spatiotemporal composite stress field.

[0051] The simulation operation is performed sequentially on each recovery path sequence in the initial strategy spectrum tree, and the distribution change of the spatiotemporal composite stress field in the power grid dynamic situation diagram is calculated after each simulation operation.

[0052] If a simulation operation causes the stress value at any location in the spatiotemporal composite stress field to exceed the safety threshold, the path branch containing the simulation operation will be pruned according to the preset safety threshold.

[0053] Specifically, to uniformly quantify the multi-dimensional physical security constraints of the power grid into a computable and comparable scalar field, a spatiotemporal composite stress field is constructed. This stress field has stress values ​​at each key equipment location in the power grid dynamic situation diagram. These stress values ​​are weighted by spatial stress weights and temporal stress weights. Spatial stress weights... This reflects the immediate risk at the moment of operation, primarily consisting of line overload risk and node voltage exceeding limits risk. Time stress weighting. This reflects potential short-term pressures, primarily composed of load increase pressures and fluctuations in renewable energy supply. For any critical location in the power grid... Its spatiotemporal composite stress value The calculation formula is:

[0054] ;

[0055] in, and The preset weighting coefficients satisfy... It is used to adjust the relative importance of current risks and future risks. It is a normalized spatial stress value, obtained by mapping real-time operating parameters such as line load rate and node voltage offset using a nonlinear function, so that its value range falls within the range of Interval. It is a normalized time stress value, quantified by analyzing the slope of the ultra-short-term load forecast curve and the new energy power generation forecast curve attached to the power grid dynamic situation diagram, and similarly processed to... To verify the physical feasibility of each recovery path sequence in the initial strategy tree, a depth-first or breadth-first traversal algorithm is used to simulate each recovery path sequence sequentially. For each operation step in the sequence, the operation is executed in a digital copy of the power grid dynamic situation diagram, such as changing the state of a switch. After the operation is completed, a full-network power flow calculation is immediately performed to obtain the voltage distribution and power flow of the entire network at the instant after the operation. Based on the power flow calculation results, the spatiotemporal composite stress field distribution at all key locations in the power grid dynamic situation diagram is updated according to the above formula. This process simulates the actual electrical response of the power grid after the dispatch command is issued. To dynamically remove unsafe recovery paths according to preset safety standards, a global scan is performed after each simulation operation and the spatiotemporal composite stress field distribution is updated to check for any points in the field where the stress value exceeds the limit. An internally preset safety threshold is used. These are engineering parameters set based on scheduling procedures and operational experience, for example, a value of 0.8. If, after a certain simulation operation, at any location in the spatiotemporal composite stress field... stress value Exceeding this safety threshold, i.e. If the simulation operation fails, it is determined that the operation will lead the power grid into an unacceptable risk state. The current path branch containing the simulated operation and all its subsequent sub-branches are pruned from the initial strategy tree, terminating further simulations of that path. Through this simulation and pruning process, the previously large and inconsistent initial strategy tree is significantly simplified. All recovery path sequences that would cause line overload, node voltage exceeding limits, or push the power grid into a future high-risk operating range during simulation execution are removed. The output is a smaller, "refined strategy tree" where all branch paths have passed preliminary safety checks at the physical level, laying a solid foundation for subsequent introduction of flexible resources for strategy evolution and in-depth optimization.

[0056] Optionally, flexible adjustment operators stored in the flexible resource operator library are invoked to generate the evolved composite strategy set, including:

[0057] During the deduction and pruning of the initial strategy phylogenetic tree, the key overload paths or voltage weak nodes that lead to pruning are identified.

[0058] From the flexible resource operator library, a target flexible adjustment operator that can alleviate the pressure on the critical overload path or the voltage-weak node is retrieved, wherein the target flexible adjustment operator includes an energy storage discharge operator and a controllable load reduction operator;

[0059] After the simulation operation steps corresponding to the critical overload path or the voltage-weak node, a target flexible adjustment operator is inserted to generate a composite strategy that integrates network reconstruction operation and flexible resource adjustment operation, thus generating an evolved composite strategy set.

[0060] Specifically, to accurately locate and quantify the physical bottlenecks leading to strategy failure, during the deduction and pruning of the initial strategy phylogenetic tree using a spatiotemporal composite stress field, whenever a path branch is pruned, it is not simply discarded, but the direct cause of the pruning is recorded. This cause is structurally recorded as a "failure event," including the operational steps that triggered the pruning, specific device identifiers such as line ID or node ID, and the physical quantity exceeding the limit, such as the line load rate reaching a certain threshold. Or the node voltage drops to These recorded devices are identified as critical overload paths or voltage-weak nodes. To intelligently match the optimal solution from available adjustment resources based on the identified bottlenecks, a targeted search request is initiated to the flexible resource operator library based on the device location and problem type recorded in the failure event. For example, if a critical overload path is identified, the search condition is "find flexible resources that can reduce the power flow of this path"; if a voltage-weak node is identified, the search condition is "find flexible resources that can provide reactive or active power support to this node and its surrounding area". Using topology information and sensitivity analysis from the power grid dynamic situation diagram, the electrical impact factor of each flexible resource on the problem point is calculated. For example, energy storage discharge operators or controllable load reduction operators located downstream of the overload path and possessing sufficient adjustable capacity will be preferentially selected. The target flexible adjustment operator is a standardized instruction object, such as {Resource ID: BESS_01, Type: Energy Storage Discharge Operator, Adjustment Amount:} Response time: To transform passively pruned paths into a composite strategy of active regulation, the pruned path sequence is retrieved, and the simulated operation step that triggered the pruning is located. The target flexible adjustment operator is inserted as a new operation step, either before or concurrently with this critical operation. This insertion is a logical "repair," intended to offset the negative impact of potential network reconfiguration operations by adjusting flexible resources before or simultaneously with them. For example, before closing a power transfer line, an instruction is inserted to command energy storage stations in the power outage area to discharge at a specific power. In this way, the original single, pure network reconfiguration operation strategy evolves into a composite strategy that integrates network reconfiguration and flexible resource adjustment operations. By performing the above identification, retrieval, and insertion steps, strategies that were originally abandoned due to failure to meet strict physical constraints are "revived" and "evolved." The output is a completely new set of strategies, namely the evolved composite strategy set. Each composite strategy in this set is a sequence of timing instructions that includes two operation types: topology switching and power regulation. It is not only topologically feasible, but also has higher security and feasibility at the physical level through proactive resource coordination, providing higher-quality candidate solutions for adversarial simulations.

[0061] Optionally, the injection of interfering factors provided by the interfering factor library for adversarial deduction includes:

[0062] In the digital twin sandbox, simulated operations are performed according to the operation sequence and parameters of the composite strategy;

[0063] At preset key operation time points or equipment status change points, several interference factors simulating instantaneous load changes, unexpected power supply failures, or communication delays are automatically injected into the simulation instance.

[0064] Monitor and record the changes in power grid state parameters of the simulation instance under the influence of the interference factor. If equipment exceeds the limit, the simulation instance is determined to have failed under the interference factor, and the cause of failure and risk migration trajectory are recorded.

[0065] Based on the success rate and state deviation of the simulation examples under different interference factor injections, the robustness index is calculated.

[0066] Specifically, for each composite strategy in the evolved composite strategy set, an independent, parallel simulation instance is created for it in the digital twin sandbox. This simulation instance is a complete clone of the power grid dynamic situation diagram at a specific moment. Then, within this instance, simulation operations are executed strictly according to the operational sequence and parameters defined by the composite strategy, such as... Always close switch A, and at the same time The system commands energy storage station B to discharge at 5 MW. This is to apply controllable, high-impact disturbances at critical nodes in strategy execution. During the simulation, the instance's status is continuously monitored. When preset critical operation time points or equipment status change points are met, the disturbance injection mechanism is automatically triggered. For example, 500 milliseconds after a large-capacity load is successfully transferred to a new line, disturbance factors are randomly or sequentially extracted from a disturbance factor library and injected into the simulation instance. These disturbance factors are standardized disturbance models, such as a 2-second load transient change model simulating the startup of a large motor in an industrial park, with peak values ​​reaching [a certain percentage] of normal load. Or, simulate a photovoltaic power station experiencing a sudden drop in power within one second due to cloud cover. The simulation includes scenarios such as unexpected disconnection of distributed power sources, or communication delays of 300 milliseconds due to control channel congestion leading to adjustments to flexible resources. To accurately capture the strategy's response to disturbances and determine its effectiveness, after the interference factor is injected, changes in key grid state parameters in the simulation examples are monitored and recorded frequently, including but not limited to node voltages, line power flow, and system frequency. These real-time changing parameters are compared with preset dynamic safety boundaries. If any equipment parameter exceeds its limit within a predetermined time window after the disturbance occurs, such as the instantaneous power flow of a line exceeding its short-term overload capacity... The simulation instance is immediately determined to have failed under the current interference factor. The entire dynamic process from interference injection to eventual failure is recorded in detail, forming a visualized risk migration trajectory that reveals the root cause of failure and the transmission path of risk. To synthesize the results of multiple adversarial simulations, a quantitative strategy robustness evaluation index is generated. Each composite strategy undergoes dozens of adversarial simulations with different interference factors. After the simulations are completed, the robustness index of the strategy is calculated based on the results of all simulation instances. The calculation formula is:

[0067] ;

[0068] in, As the final robustness indicator; It is the number of simulation instances that did not fail in all adversarial simulations; This is the total number of injected interfering factors, i.e., the total number of tests. This represents the success rate of the strategy; This is a coefficient reflecting the stability of successful cases. Its value is determined by the average of the maximum deviations from the grid state in successful cases. The smaller the deviation, the higher the stability. The closer the value is to 1, the smaller it is, thus penalizing strategies that, while successful, approach the safety margin. By executing the aforementioned adversarial simulation process, a detailed "health check report" is generated for each candidate composite strategy. The core of this report is a quantitative robustness indicator, intuitively reflecting the strategy's ability to withstand unknown risks. Furthermore, the report includes risk migration trajectories for all failure cases. This trajectory data provides decision-makers with valuable insights into the potential weaknesses and risks of the strategy, offering crucial resilience-level decision-making basis for the final multi-level game evaluation.

[0069] Optionally, the method further includes:

[0070] Collect the deduction process data and result data of all the aforementioned deduction examples to form a strategy deduction case library;

[0071] Analyze the strategy gene combination patterns of successful cases in the strategy deduction case library and the corresponding power grid dynamic situation map characteristics to strengthen the correlation weight of the strategy gene operator in the digital contingency plan library;

[0072] Analyze the failure causes of failed cases and extract the failure causes into new constraints, which can be used to optimize the calculation model of spatiotemporal composite stress field or enrich the interference factor library.

[0073] Specifically, to construct a structured, analyzable database of strategic experience, complete data from all simulation examples is automatically collected and archived. This includes detailed operational sequences for each composite strategy, the power grid dynamic situation diagram used before the simulation, the types and parameters of injected interference factors, time-series data of all key electrical quantities during the simulation, risk migration trajectories, and the final simulation results. This data is integrated, labeled, and stored in a strategy simulation case library. This library is a large-scale, multi-dimensional database, providing raw materials for subsequent knowledge extraction and model optimization. To extract and strengthen effective strategy patterns from successful cases, a background analysis task is initiated to periodically perform data mining on successful cases in the strategy simulation case library. This task applies association rule mining algorithms to analyze which combinations of strategy gene operators exhibit the highest success rate and robustness under specific power grid dynamic situation diagram characteristics. Once such a strong correlation pattern is discovered, the association weight of the strategy gene operators constituting that pattern in the digital contingency plan library is automatically strengthened. The increased weighting means that when encountering similar power grid situations in the future, policy branches containing combinations of these efficient operators will be given higher priority in the initial policy phylogenetic tree generation stage, thereby accelerating the discovery of high-quality policies. To learn from failed cases and fill cognitive gaps in the model, a special analysis is conducted on failed cases in the policy simulation case library, particularly focusing on in-depth analysis of recorded failure causes and risk migration trajectories. Root cause analysis can identify potential risks that the current model has not fully considered. For example, the analysis may reveal that a certain type of composite policy is highly likely to cause coordination failure and oscillations when communication latency exceeds 500 milliseconds. This newly discovered failure mechanism will be refined into a new constraint. Based on this constraint, the calculation model of the spatiotemporal composite stress field will be modified, for example, by adding a penalty term related to communication status, so that the model can identify and avoid such risks in advance during the simulation pruning stage. Alternatively, if the failure is found to be caused by an unexpected combination of disturbances, this disturbance scenario will be modeled, refined into a new disturbance factor, and added to the disturbance factor library to enhance the system's test completeness in future adversarial simulation stages. By implementing this closed-loop learning mechanism, a dynamic and self-improving decision-making knowledge system is constructed. The continuous accumulation of the strategy deduction case library provides the system with nourishment for continuous learning. The analysis of successful cases makes the system's strategy generation process more "skilled" and efficient; the analysis of failed cases continuously strengthens the system's safety baseline, enabling it to identify and respond to more and more complex unknown risks. This process ensures that the decision support method not only provides support in a single event, but also continuously evolves with the growth of operational experience, improving its long-term decision-making intelligence level.

[0074] Optionally, the application uses a pre-defined multi-level game evaluation model comprising a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate the various composite strategies used in the adversarial simulation, including:

[0075] The first layer of security evaluation filters out composite strategies that do not cause any device parameter exceedances in the adversarial simulation, forming a set of security strategies;

[0076] The second efficiency evaluation layer calculates the power restoration ratio and the total number of operation steps for each composite strategy in the set of security strategies, and performs an efficiency trade-off.

[0077] The third resilience layer evaluation assesses the resilience of candidate strategies after the efficiency layer evaluation based on the robustness index and the dispersion of the risk migration trajectory.

[0078] Based on the combined results of the safety layer evaluation, efficiency layer evaluation, and resilience layer evaluation, the Pareto optimal frontier method is used to select the recommended strategy.

[0079] Specifically, the first layer is the security layer evaluation, which conducts a basic security qualification review and unconditionally eliminates any strategies with known risks. The system first traverses all composite strategies that have completed adversarial simulations. For each strategy, its performance in all simulation instances is checked, including performance under both injected and non-injected interference factors. The screening criteria are extremely strict; only composite strategies that do not cause any equipment parameter exceedances in any simulation from start to finish are considered to have passed the security layer evaluation. Any strategy with even one simulation failure record is directly eliminated. Through this level of filtering, a set of strategies that perform perfectly in the simulation environment is obtained, namely the safe strategy set. The second layer is the efficiency layer evaluation, which quantitatively compares the economy and speed of the strategies while ensuring safety. For each composite strategy in the safe strategy set, two core efficiency indicators are calculated. The first is the power restoration ratio, which is the ratio of the total load restored after the strategy is executed to the total load lost before the fault. The second is the total number of operation steps, i.e., the total number of switching operations and flexible resource adjustment instructions included in the strategy, which directly relates to the complexity of the operation and the execution time. An efficiency trade-off is made between these two inherently conflicting metrics. Generally, a higher power restoration rate is better, while a lower total number of operation steps is better. The goal of this evaluation layer is not to directly provide a single optimal solution, but to identify advantageous strategies under different efficiency preferences. The third layer is the resilience layer evaluation, which assesses the robustness and risk controllability of the strategy in the face of unknown disturbances; this is a measure of high-level security performance. Using data obtained during the adversarial simulation phase, the resilience of candidate strategies after the efficiency layer evaluation is assessed. The evaluation mainly relies on two parameters: first, the robustness index, which was calculated during the adversarial simulation phase and directly reflects the success rate and stability of the strategy against various disturbances; and second, the dispersion of risk migration trajectories. Analyzing the risk migration trajectories of all failed scenarios for a strategy, if these trajectories are highly concentrated, pointing to a few predictable failure points, it indicates that the risk is convergent and relatively controllable; conversely, if the trajectories are highly divergent, it indicates that the strategy may trigger a variety of unpredictable cascading failures under different perturbations, with high risk dispersion and poor resilience. Resilience assessment tends to select strategies with high robustness indices and concentrated risk migration trajectories. To find one or a set of non-dominated solutions without absolute superiority or inferiority relationships across the three mutually constraining objective dimensions of safety, efficiency, and resilience, the evaluation results of the above three levels are synthesized, and each candidate strategy is mapped to a multi-dimensional objective space composed of power restoration ratio, total number of operation steps, robustness indices, etc. Then, the Pareto optimal front method is used for solution. A strategy that is not inferior to another strategy in all evaluation dimensions and is strictly superior to it in at least one dimension is considered to dominate the latter. All strategies not dominated by any other strategy constitute the Pareto optimal front.This set of recommended strategies at the forefront represents all possible "optimal" trade-offs under the current constraints. By implementing a multi-level game evaluation model, the output is no longer a single "best" answer, but a set of Pareto-optimal recommended strategies that have undergone rigorous security checks and exhibit different advantages in terms of efficiency and resilience. This provides scheduling decision-makers with a high-quality menu of options, allowing them to make a final, practically necessary decision from this set of recommended strategies based on current scheduling objectives.

[0080] Optionally, the process of selecting a recommendation strategy further includes:

[0081] The operation sequence and flexible adjustment instructions corresponding to the recommended strategy are converted into structured operation tickets that conform to the scheduling procedure.

[0082] The structured operation ticket, the risk migration trajectory, and key early warning information are visualized.

[0083] Specifically, to achieve automated translation from policy models to scheduling instructions, the process involves obtaining the final selected recommended policy, which is essentially an ordered list of policy gene operators and flexible adjustment operators. The instruction conversion engine is then activated, which has a built-in library of syntax rules and templates. The engine iterates through each operator in the policy, automatically populating the structured information such as the execution subject and action instructions into a standard operation ticket template. For example, a policy gene operator {execution subject: line 952 switch, action instruction: close} will be converted to "operation task: close". Sunshine Substation "Outgoing line 952 switch". For the flexible adjustment operator, control commands conforming to a specific resource communication protocol will be generated, such as {Resource ID: BESS_01, Action: Discharge, Parameter: Power}. Duration Ultimately, the entire strategy sequence is transformed into one or more logically rigorous, formatted, and directly executable structured operation tickets. This provides decision-makers with in-depth insights and forward-looking warnings regarding strategy execution risks. All adversarial simulation data associated with the recommended strategy are retrieved, especially those risk migration trajectories that record the failure process. These trajectories are dynamically displayed on the power grid's geographical wiring diagram or topology map using a data visualization engine. For example, using flashing arrows of different colors and heat maps, the entire process of how the power flow overload risk in the power grid gradually propagates and amplifies from weak points after the injection of a certain interference factor, ultimately leading to another remote device exceeding its limits is vividly illustrated. Simultaneously, key warning information is extracted from these trajectories, such as "After executing step 3, if the load growth in area A exceeds..." Line 734 exists The system identifies and highlights the "overload risk" along with structured operation tickets. This visualization allows dispatchers to intuitively understand the potential weaknesses of the strategy and its dynamic response under specific disturbances, enabling them to be well-prepared and proactive in actual operations. By executing these two steps, not only is a thoroughly refined "optimal" strategy provided, but also a complete set of easy-to-understand, easy-to-execute, and risk-transparent decision support tools. The output structured operation tickets greatly reduce the burden on dispatchers in compiling operation instructions and lower the risk of human error. The visualization of risk migration trajectories and key early warning information bridges the gap between decision-makers and complex simulation results, transforming abstract robustness indicators into concrete and perceptible risk scenarios, thereby enhancing decision-makers' trust in the recommended strategy and their risk management capabilities during actual execution.

[0084] Based on the same inventive concept, such as Figure 3 As shown, the present invention also provides a rapid power transfer auxiliary decision-making system based on digital contingency plans and topology simulation, the system comprising:

[0085] The data acquisition and identification module is used to collect real-time operation data of the power grid, environmental prediction data, and flexible resource status data, and to identify power grid fault information based on the real-time operation data.

[0086] The situation construction and sand table instantiation module is used to integrate the power grid fault information, the environmental prediction data and the flexible resource status data to construct a dynamic situation map of the power grid, which includes the power grid topology connection relationship, the real-time and future predicted carrying capacity parameters of each path and the callable flexible adjustment capacity parameters of each node. Based on the dynamic situation map of the power grid, a digital twin sand table that runs synchronously with the physical power grid is constructed in the digital twin environment.

[0087] The contingency plan recombination and phylogenetic tree generation module is used to obtain strategy gene operators that match the power grid fault information from the digital contingency plan library constructed based on text contingency plan parsing, and dynamically recombine the strategy gene operators according to the characteristics of the power grid dynamic situation map to generate an initial strategy phylogenetic tree.

[0088] The stress field deduction and composite evolution module is used to map the physical security constraints in the power grid dynamic situation diagram into a spatiotemporal composite stress field, use the spatiotemporal composite stress field to deduce and prune the initial strategy phylogenetic tree, and call the flexible adjustment operators stored in the flexible resource operator library to generate the evolved composite strategy set.

[0089] The sand table adversarial simulation and indicator acquisition module is used to create parallel simulation instances for each composite strategy in the composite strategy set in the digital twin sand table, inject interference factors provided by the interference factor library to conduct adversarial simulation, and obtain the risk migration trajectory and robustness indicators of each composite strategy.

[0090] The game evaluation and strategy recommendation module is used to apply a pre-set multi-level game evaluation model that includes a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate the various composite strategies that have completed adversarial simulations, and output recommended strategies based on the evaluation results.

[0091] Example 1:

[0092] To verify the effectiveness of the method described in this invention, this embodiment illustrates a typical fault scenario of a city's power distribution network. In this scenario, One of the substations of Chaoyang (hereinafter referred to as "Chaoyang Station") Outgoing line L952 tripped due to a cable fault, causing a complete power outage in area A, which it supplies. The power outage load is... This includes critical users such as hospitals and data centers. The goal of this method is to enable dispatchers to quickly generate secure, efficient, and reliable power restoration strategies.

[0093] First, real-time operational data is obtained from the power grid energy management system (EMS), and fault information is identified. Simultaneously, environmental forecast data and flexible resource status data are retrieved. Specific parameters are as follows: Fault information: Chaoyang Station Line L952 switch tripped, resulting in power loss of load. Real-time operating data: The current load of the adjacent backup line L954 is... Its rated transmission capacity is The other backup line, L956, currently has a load of 4MW and a rated capacity of [missing information]. However, the line is relatively long, and the voltage support at the end is weak. Environmental forecast data: The meteorological system predicts high temperatures within the next 2 hours, and the regional load is expected to rise across the board. Flexible resource status data: An energy storage station ES-01 is configured in area A, with a current available capacity of... State of charge (SOC) is The maximum discharge power is Simultaneously, an agreement was signed with a commercial building aggregator, granting access to the controllable air conditioning load CL-01, which can be reduced at any time. .

[0094] By integrating the above data, a dynamic power grid status map is constructed. This map not only includes feasible physical power transfer paths such as L954 and L956, but also marks the predicted carrying capacity parameters of L954, as well as the callable flexible regulation capacity parameters of ES-01 and CL-01.

[0095] Based on the fault information "L952 line tripped", the basic operation operator sequence is indexed from the digital contingency plan library. This sequence is usually a standard operation based on historical experience, namely "transfer power through the backup line L954". Using this as the root node, and based on the topology connection relationship in the dynamic situation diagram, other logically equivalent power transfer paths are searched to generate an initial strategy genealogy tree containing multiple branches.

[0096] Table 1 Initial Strategy Lineage Tree Generation

[0097] Branch number recovery path Strategy Gene Operator Sequence illustrate Branch 1 Via L954 {Close the L954 communication switch} Standard contingency plan path Branch 2 Via L956 {Close the L956 communication switch} Topological Equivalent Path

[0098] Physical security constraints are mapped to a spatiotemporal composite stress field, and simulations are performed on each branch of the initial strategy phylogenetic tree. The stress value calculation formula is as follows: Safety threshold Set to 0.8. Deduction Branch 1 (via L954): Simulation Operation: Close the L954 interconnect switch in the dynamic situation diagram copy. Power Flow Calculation: The total load of L954 becomes... The line load rate reached It is already in an overloaded state. Spatial stress. Due to severe overload, the normalized line overload risk The calculated value is 0.95. Time stress. Considering the future The load increase, the original The load will increase to The total load will then reach The overload situation will worsen further. Quantized time stress weights. The value is 0.7. Composite stress calculation: Pruning decisions: due to The path was deemed to have an unacceptable risk, so branch 1 was pruned and the failure reason was recorded as "line L954 overload".

[0099] Deduction Branch 2 (via L956): Simulation Operation: Close the L956 tie switch. Power Flow Calculation: The total load of L956 is... Load rate No overload. However, due to the long line length, the voltage at the end node N10 in area A increased from... Descending to Spatial stress Node voltage is lower than Safety lower limit, normalized voltage over-limit risk The calculated value is 0.9. Composite stress calculation: The spatial stress component alone has made it highly likely that the total stress value will exceed the limit. Pruning decision: Determine that the path will lead to a serious voltage problem, prune branch 2, and record "node N10 voltage exceeds limit".

[0100] All branches of the initial strategy phylogenetic tree were pruned. The stress field extrapolation and composite evolution module was activated to repair the pruned paths. Repairing branch 1 (L954 overload problem): Identifying the bottleneck: the critical overload path L954. Retrieving flexible resources: Resources that can be used to reduce the L954 power flow were found, namely the dischargeable energy storage station ES-01 located in area A. The controllable load CL-01 can reduce Generate composite strategy A1: Before closing the L954 switch, first invoke flexible resources. The new operation sequence is: {Instruction CL-01 Reduce load} Command ES-01 to discharge Close the L954 interconnecting switch. Verification: After flexible resource adjustment, the actual load that needs to be transferred from L954 is reduced to... The total load of L954 becomes Load rate The total load after future load growth will be... Short-term overload is within acceptable limits. This composite strategy has passed safety verification.

[0101] Repairing Branch 2 (N10 Voltage Issue): Bottleneck identified: Voltage-weak node N10. Flexible resource retrieval: Energy storage station ES-01 is found to be electrically located near N10; its inverter can provide reactive power support while discharging. Generate composite strategy B1: {Instruct ES-01 to...} meritorious and In reactive power mode operation, close the L956 tie switch. Verification: Power flow calculations show that... The reactive power injection can boost the voltage of N10 to The system is then restored to a safe range. The composite strategy passes the safety check. This generates an evolved composite strategy set {A1, B1} containing two feasible paths.

[0102] Parallel simulation instances of A1 and B1 were created in the digital twin sandbox, and interference factors were injected for stress testing.

[0103] Table 2. Results of the adversarial simulation and evaluation

[0104] Strategy Injecting interference factors Deduction results Robustness index R efficiency indicators Resilience Index A1 1. Load surge of 1MW in Area A 2. ES-01 communication delay of 500ms 1. L954 experienced a momentary overload of 105%, but did not trip. 2. L954 experienced a severe momentary overload, causing a voltage drop in N10, resulting in a power outage at the data center (simulation failed). 0.75 5MW / 3 steps Risk trajectories are dispersed and sensitive to communication delays. B1 1. Load surge of 1MW in Area A 2. ES-01 communication delay of 500ms 1. L956 load rate 98%, voltage stable. 2. Voltage experienced a slight momentary dip but recovered quickly. 0.95 5MW / 2 steps The risk trajectory converges and is insensitive to both load and latency.

[0105] Security Layer Evaluation: Both strategies A1 and B1 are safe under no-interference conditions and are included in the safe strategy set. Efficiency Layer Evaluation: Strategy A1 requires 3 steps, while B1 requires only 2 steps, and both restore the same load. B1 is more efficient. Resilience Layer Evaluation: Strategy B1's robustness index (0.95) is significantly higher than A1's (0.75), and its risk migration trajectory is more convergent, indicating that it performs more stably and reliably under uncertain disturbances.

[0106] Table 3 Final Strategy Recommendations

[0107] Recommended strategy Overall evaluation Recommendation reason Strategy B1 Pareto optimal solution Under the premise of restoring the same load, fewer operation steps are required, and it exhibits higher robustness in adversarial simulations, resulting in the best overall performance. Strategy A1 Suboptimal solution Although feasible, it involves many steps and carries a high risk under certain interferences such as communication delays.

[0108] Finally, strategy B1 is recommended to the dispatcher, and a structured operation ticket is automatically generated: "Instruction A area ES-01..." meritorious and Run in reactive mode; after confirmation, close. Line L956 interconnecting switch. For example... Figure 4 As shown, the distribution of each candidate composite strategy in terms of power restoration efficiency and robustness is displayed through a multi-dimensional objective space. The solid lines connecting the strategies represent the Pareto optimal frontier, reflecting the superior position of recommended strategy B1 in the comprehensive evaluation. Simultaneously, a visualization interface is used to display the risk migration trajectory of strategy A1 under communication delays, serving as auxiliary decision-making information. This embodiment fully demonstrates how the present invention, through a series of interconnected derivations and calculations, can quickly and scientifically generate optimal recovery strategies that combine safety, efficiency, and resilience from complex power grid fault scenarios.

[0109] All equivalent changes and modifications made in accordance with the teachings of this invention shall still fall within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or customary techniques in the art not described herein.

Claims

1. A rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation, characterized in that, The method includes: Collect real-time operation data of the power grid, environmental prediction data, and flexible resource status data, and identify power grid fault information based on the real-time operation data; By integrating the power grid fault information, the environmental prediction data, and the flexible resource status data, a dynamic power grid situation map is constructed, which includes the power grid topology connection relationship, the real-time and future predicted carrying capacity parameters of each path, and the callable flexible adjustment capacity parameters of each node. Based on the dynamic power grid situation map, a digital twin sandbox that runs synchronously with the physical power grid is constructed in the digital twin environment. The strategy gene operator that matches the power grid fault information is obtained from the digital contingency plan library constructed based on text contingency plan parsing, and the strategy gene operator is dynamically recombined according to the characteristics of the power grid dynamic situation map to generate an initial strategy genealogy tree. The physical security constraints in the power grid dynamic situation diagram are mapped to a spatiotemporal composite stress field. The spatiotemporal composite stress field is used to deduce and prune the initial strategy phylogenetic tree, and the flexible adjustment operators stored in the flexible resource operator library are called to generate the evolved composite strategy set. In the digital twin sandbox, a parallel simulation instance is created for each composite strategy in the composite strategy set, and interference factors provided by the interference factor library are injected to conduct adversarial simulation to obtain the risk migration trajectory and robustness index of each composite strategy. A pre-defined multi-level game evaluation model, comprising a security layer, an efficiency layer, and a resilience layer, is used to comprehensively evaluate the various composite strategies that complete the adversarial simulation, and recommended strategies are output based on the evaluation results.

2. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction as described in claim 1, characterized in that, By integrating the power grid fault information, the environmental prediction data, and the flexible resource status data, a dynamic power grid situation map is constructed, including power grid topology connections, real-time and future predicted carrying capacity parameters of each path, and callable flexible adjustment capacity parameters of each node. Based on the power grid topology and power flow information in the real-time operating data, a current power grid topology connection diagram is generated; The ultra-short-term load forecast curve and the new energy power generation forecast curve in the environmental forecast data are mapped to the relevant equipment and paths in the current power grid topology connection diagram to calculate the predicted carrying capacity parameters for several predetermined periods in the future. The status data of the flexible resources are obtained, and the adjustable capacity parameters of each flexible resource are marked on the corresponding nodes of the current power grid topology connection diagram to form a dynamic status diagram of the power grid.

3. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction as described in claim 1, characterized in that, Based on the characteristics of the power grid dynamic situation map, the strategy gene operator is dynamically recombined to generate an initial strategy phylogenetic tree containing several recovery path sequences, including: Natural language processing is performed on the text plan to parse it into a structured strategy gene operator consisting of triggering conditions, execution subject, action instructions and security boundaries, and then stored in the digital plan library. Based on the fault equipment identifier and power outage range in the power grid fault information, the basic operation operator sequence is indexed from the digital contingency plan database; Using the basic operation operator sequence as the root node, and based on the real-time carrying capacity parameters and connection relationships of each path in the power grid dynamic situation diagram, the operation objects and order in the basic operation operator sequence are logically equivalently replaced and expanded to form several branches, thus constituting the initial strategy spectrum tree.

4. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction as described in claim 1, characterized in that, The deduction and pruning of the initial strategy phylogenetic tree using the spatiotemporal composite stress field includes: The overload risk of the line and the voltage over-limit risk of the node are quantified into spatial stress weights, and the load increase pressure and new energy fluctuation pressure in the future predetermined period are quantified into temporal stress weights, which together constitute a spatiotemporal composite stress field. The simulation operation is performed sequentially on each recovery path sequence in the initial strategy spectrum tree, and the distribution change of the spatiotemporal composite stress field in the power grid dynamic situation diagram is calculated after each simulation operation. If a simulation operation causes the stress value at any location in the spatiotemporal composite stress field to exceed the safety threshold, the path branch containing the simulation operation will be pruned according to the preset safety threshold.

5. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction as described in claim 1, characterized in that, The flexible adjustment operators stored in the flexible resource operator library are invoked to generate the evolved composite policy set, which includes: During the deduction and pruning of the initial strategy phylogenetic tree, the key overload paths or voltage weak nodes that lead to pruning are identified. From the flexible resource operator library, a target flexible adjustment operator that can alleviate the pressure on the critical overload path or the voltage-weak node is retrieved, wherein the target flexible adjustment operator includes an energy storage discharge operator and a controllable load reduction operator; After the simulation operation steps corresponding to the critical overload path or the voltage-weak node, a target flexible adjustment operator is inserted to generate a composite strategy that integrates network reconstruction operation and flexible resource adjustment operation, thus generating an evolved composite strategy set.

6. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction according to claim 1, characterized in that, The injection of interference factors provided by the interference factor library for adversarial deduction includes: In the digital twin sandbox, simulated operations are performed according to the operation sequence and parameters of the composite strategy; At preset key operation time points or equipment status change points, several interference factors simulating instantaneous load changes, unexpected power supply failures, or communication delays are automatically injected into the simulation instance. Monitor and record the changes in power grid state parameters of the simulation instance under the influence of the interference factor. If equipment exceeds the limit, the simulation instance is determined to have failed under the interference factor, and the cause of failure and risk migration trajectory are recorded. Based on the success rate and state deviation of the simulation examples under different interference factor injections, the robustness index is calculated.

7. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction according to claim 6, characterized in that, The method further includes: Collect the deduction process data and result data of all the aforementioned deduction examples to form a strategy deduction case library; Analyze the strategy gene combination patterns of successful cases in the strategy deduction case library and the corresponding power grid dynamic situation map characteristics to strengthen the correlation weight of the strategy gene operator in the digital contingency plan library; Analyze the failure causes of failed cases and extract the failure causes into new constraints, which can be used to optimize the calculation model of spatiotemporal composite stress field or enrich the interference factor library.

8. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction according to claim 1, characterized in that, The application uses a pre-defined multi-level game evaluation model that includes a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate the various composite strategies that complete the adversarial simulation, including: The first layer of security evaluation filters out composite strategies that do not cause any device parameter exceedances in the adversarial simulation, forming a set of security strategies; The second efficiency evaluation layer calculates the power restoration ratio and the total number of operation steps for each composite strategy in the set of security strategies, and performs an efficiency trade-off. The third resilience layer evaluation assesses the resilience of candidate strategies after the efficiency layer evaluation based on the robustness index and the dispersion of the risk migration trajectory. Based on the combined results of the safety layer evaluation, efficiency layer evaluation, and resilience layer evaluation, the Pareto optimal frontier method is used to select the recommended strategy.

9. The rapid power transfer auxiliary decision-making method based on digital contingency plans and topology deduction as described in claim 8, characterized in that, Following the selection and recommendation strategy, the following also includes: The operation sequence and flexible adjustment instructions corresponding to the recommended strategy are converted into structured operation tickets that conform to the scheduling procedure. The structured operation ticket, the risk migration trajectory, and key early warning information are visualized.

10. A rapid power transfer auxiliary decision-making system based on digital contingency plans and topology simulation, characterized in that: The system is used in the rapid power transfer auxiliary decision-making method based on digital contingency plans and topology simulation as described in any one of claims 1-9, and the system comprises: The data acquisition and identification module is used to collect real-time operation data of the power grid, environmental prediction data, and flexible resource status data, and to identify power grid fault information based on the real-time operation data. The situation construction and sand table instantiation module is used to integrate the power grid fault information, the environmental prediction data and the flexible resource status data to construct a dynamic situation map of the power grid, which includes the power grid topology connection relationship, the real-time and future predicted carrying capacity parameters of each path and the callable flexible adjustment capacity parameters of each node. Based on the dynamic situation map of the power grid, a digital twin sand table that runs synchronously with the physical power grid is constructed in the digital twin environment. The contingency plan recombination and phylogenetic tree generation module is used to obtain strategy gene operators that match the power grid fault information from the digital contingency plan library constructed based on text contingency plan parsing, and dynamically recombine the strategy gene operators according to the characteristics of the power grid dynamic situation map to generate an initial strategy phylogenetic tree. The stress field deduction and composite evolution module is used to map the physical security constraints in the power grid dynamic situation diagram into a spatiotemporal composite stress field, use the spatiotemporal composite stress field to deduce and prune the initial strategy phylogenetic tree, and call the flexible adjustment operators stored in the flexible resource operator library to generate the evolved composite strategy set. The sand table adversarial simulation and indicator acquisition module is used to create parallel simulation instances for each composite strategy in the composite strategy set in the digital twin sand table, inject interference factors provided by the interference factor library to conduct adversarial simulation, and obtain the risk migration trajectory and robustness indicators of each composite strategy. The game evaluation and strategy recommendation module is used to apply a pre-set multi-level game evaluation model that includes a security layer, an efficiency layer, and a resilience layer to comprehensively evaluate the various composite strategies that have completed adversarial simulations, and output recommended strategies based on the evaluation results.

Images